Brazil: the Supreme Electoral Court wiped out voters’ registrations

By Ryan Lloyd (University of São Paulo), GLOBALCIT collaborator


On October 7, 3.3 million voters received an unwelcome surprise on Election Day in Brazil. They were, in fact, no longer voters.

Roughly half (2,793) of the country’s 5,570 municipalities had moved to compulsory biometric identification for this election, which required voters to go to the Electoral Court in their region and register their fingerprints. Unfortunately, as sometimes happens with major bureaucratic changes, the communication of this change was less than ideal, and 41% of the electorate never did make this trip; of this segment, 3,368,000 voters, or 2.3% of the electorate (an increase in cancellations of 183% from 2014), lived in areas in which doing so was compulsory, and duly had their electoral registration cancelled.

This left the Supreme Electoral Court (TSE) in a pickle: undermine the effort and let those who didn’t register vote anyway? Or, despite the fact that voting is compulsory, follow through and ban those who had not registered from voting?

They ultimately decided on the latter, citing the fact that voters had been given time to appeal any removal from the rolls, and arguing that it would not disproportionately affect the poor. Unfortunately, the communication of this decision was also less than ideal, as voters who had their registration canceled were not individually notified. In fact, if they hadn’t gone out of their way to check, they only found out after they showed up at the polls to vote. To make matters worse, a rumor on Whatsapp circulated claiming that you could vote even without having done the registration if you showed up with photo identification.

Voter registration was a big flashpoint in the U.S., possibly affecting the turnout of certain segments of the population disproportionately. Might it have had an effect on the Brazilian election as well? After all, any possible effect would not have been captured by polls, since voters with cancelled registration would likely have not known and responded to polls as likely voters. This would mean that our pre-electoral predictions would have been off-base if there truly was an influence.

Why voter de-registration could have had an effect

It’s difficult to know whether this was the case; as a matter of fact, we’ll likely never know for sure because the TSE does not make individual-level voting data available. We also don’t know with any certainty what the votes of those who had their registration canceled would have been had they been allowed to vote.

That said, as El País shows, more than half of the cancelled electoral registrations were in municipalities with less than 50,000 residents. These cities, taken together as a whole, were more rural than the average Brazilian municipality, with lower levels of human development, more than twice the level of extreme poverty. In fact, 1 million registrations alone were cancelled in municipalities with 30% of the population living in rural zones (twice the national average). On the one hand, the Northeast was hit especially hard, with 1.5 million registrations (44.7% of the total) being cancelled in that region alone (3.83% of the electorate). In the Southeast, on the other hand, only 1.1% of the electorate (about 700,000) had their registrations canceled.

This has possibly hurt Fernando Haddad of the center-left PT (Workers’ Party) more than anyone else. In pre-election polling, Haddad was polling especially well in the Northeast and among poorer voters. In theory, these would be the voters most likely to be taken off the voting rolls.

Why it probably didn’t have much effect

The sheer scale of Haddad’s defeat, however, makes it unlikely that the TSE’s decision had much to do with the eventual outcome. Jair Bolsonaro of the far-right PSL (Social Liberal Party) won the first round with 46.03%, narrowly missing out on the first-round majority that would have dispensed with the necessity of a runoff. Haddad trailed with 29.28%, with the PDT’s (Democratic Labor Party) Ciro Gomes finishing in third with 12.47%. In the second round, on 28 October, Bolsonaro won with a comfortable margin of 55.1% of the vote.

Furthermore, despite the scale of the defeat, Haddad did not actually underperform. After all, Brazil had a crowded field of 13 presidential candidates in the first round, not just two; in fact, both Bolsonaro and Haddad outperformed their polling figures, and they did so at the expense of other candidates such as the PSDB’s (Brazilian Social Democracy Party) Geraldo Alckmin and the Rede’s (Sustainability Network) Marina Silva, both of whom undershot their expected vote totals. Neither of them, however, had strong bases among the segments of the electorate that would have been most affected by the TSE’s decision.

It is also possible that many of those who did not register their fingerprints would never have gotten to the polls anyway even if they had been registered. Voting is compulsory, with those who do not do so incurring penalties. That, however, does not mean that everyone does vote. Voting rates are indeed high, at 79.67% of all registered voters in 2018 and 80.61% in 2014 for the first round of the presidential election, but that still leaves about 20% of all registered voters who do not actually vote. This percentage might not have voted, then, even if they had remained on the rolls.

This is not to say that Haddad (or Alckmin or Silva) might not have done even better without voter registration. Just because Haddad outperformed expectations does not mean that he could not have outperformed them by even more had voters not been barred from registering. But it is extremely unlikely that this would have happened on the scale that could have overturned a a 10-percentage point difference between Bolsonaro and Haddad in the second round, much less a 17- percentage point difference between Bolsonaro and Haddad (or Haddad and Gomes) in the first round.

In other words, we cannot, with the data at hand, know how many votes were lost and for whom in this election. But it would seem extremely unlikely to have made a difference.